Drosophila follicle cell amplicons as models for metazoan DNA replication: A cyclinE mutant exhibits increased replication fork elongation

Abstract

Gene clusters amplified in the ovarian follicle cells of Drosophila serve as powerful models for metazoan DNA replication. In response to developmental signals, specific genomic regions undergo amplification by repeated firing of replication origins and bidirectional movement of replication forks for ≈50 kb in each direction. Previous work focused on initiation of amplification, defining replication origins, establishing the role of the prereplication complex and origin recognition complex (ORC), and uncovering regulatory functions for the Myb, E2F1, and Rb transcription factors. Here, we exploit follicle cell amplification to investigate the control of DNA replication fork progression and termination, poorly understood processes in metazoans. We identified a mutant in which, during gene amplification, the replication forks move twice as far from the origin compared with wild type. This phenotype is the result of an amino acid substitution mutation in the cyclinE gene, cyclinE^1f36. The rate of oogenesis is normal in cyclinE^1f36/cyclinE^Pz8 mutant ovaries, indicating that increased replication fork progression is due to increased replication fork speed, possibly from increased processivity. The increased amplification domains observed in the mutant imply that there are not replication fork barriers preventing replication forks from progressing beyond the normal 100-kb amplified region. These results reveal a previously unrecognized role for CyclinE in controlling replication fork movement.

During follicle cell gene amplification, replication occurs by repeated firing of defined origins of replication using the same DNA replication machinery as in the normal cell cycle (for review, see ref. 1). Replication forks progress bidirectionally from these origins to produce gradients of amplified DNA in an onion-skin structure. There are at least four Drosophila amplicons in follicle cells (DAFCs) (DAFC-7F, DAFC-30B, DAFC-62D, and DAFC-66D) containing genes encoding eggshell constituents (2). Because the increased gene copy number produced by amplification is needed for adequate production of these proteins, mutants affecting gene amplification are female sterile and exhibit egg shell defects (1).

Drosophila Amplicons as Metazoan Replication Models.

Recovery of amplification mutants showed that replication proteins normally used in S phase are critical for gene amplification. Weak alleles of the essential genes for normal DNA replication, mcm6, dup/Cdt1, orc2, and chiffon, which shows homology to dbf4, a specificity factor for the S-phase kinase Cdc7, are female sterile and reduce gene amplification (3–6). It is possible to visualize DNA replication during follicle cell gene amplification directly, because it occurs after the cessation of genomic replication (7). Thus immunolocalization of replication proteins and incorporation of BrdU is observed solely at the DAFC sites of amplification. The replication factors Orc1, Orc2, double parked (DUP)/Cdt1, Cdc45, the MCM complex, and proliferating cell nuclear antigen (PCNA) localize to amplicons (3–6, 8–11).

The control elements for DAFC-66D have been defined by transformation experiments in which transposons containing fragments from DAFC-66D inserted at ectopic sites were tested for their ability to amplify with proper developmental timing. These studies defined ACE3, amplification control element from the third chromosome, as a crucial regulatory element and delineated it to 320 bp (12–14). Two dimensional gel analysis mapped the predominant replication origin, ori-β, to a region ≈1.5 kb away from ACE3 (15, 16). Using transposons buffered from inhibitory position effects by chromatin insulators, Tower and his students demonstrated that ACE3 and ori-β are the minimal elements necessary for proper developmental amplification and that ACE3 acts as a replication enhancer (17–19). ACE3 has been shown both in vivo and in vitro to bind ORC, the eukaryotic origin recognition complex (ORC) (8, 11, 20), and it seems to act to recruit ORC to localize to ori-β (19). ACE3 contains binding sites for the Myb protein complex, which regulates its amplification activity (21–23). The Rb and E2F1 transcription factors also control amplification through their presence in a complex with ORC (24).

The levels of CyclinE protein affect follicle cell amplification. Overexpression of the cyclinE gene inhibits amplification, and overexpression of the Kip-p27 like inhibitor dap also inhibits amplification (7). Thus it seems that a critical window of CyclinE activity is required for amplification. During the endo cycles that precede amplification in the follicle cells, levels of CyclinE protein oscillate but become constitutively high during amplification. It has been proposed that high CyclinE levels restrict genomic replication at all sites except the DAFCs (7). Delineation of the role of CyclinE in control of amplification, however, has been hampered by the absence of cyclinE alleles with sufficient activity to produce viable adults that support oogenesis to stages to analyze amplification.

Regulation of Replication Fork Progression.

Drosophila follicle cell amplicons have been used largely to investigate the control of replication initiation, but they also provide the opportunity to understand replication elongation in metazoans. Regulation of replication fork progression ensures proper duplication of the genome in normal S phase and in response to unexpected challenges faced by the cell (for review, see ref. 25). When DNA is damaged, replication forks stall to permit DNA repair (26). In eukaryotes there are examples of specific sites, replication fork barriers (RFBs), that block replication fork progression opposing transcription within the rDNA loci (for review, see ref. 27). Although replication termination sites have been defined in Escherichia coli and Bacillus subtilis that may facilitate decatenation of these circular chromosomes (28), it seems rare for specific replication termination sites to occur in eukaryotic DNA replication (29).

Replication fork progression is regulated, in part, by factors found at the replication fork. PCNA, the DNA polymerase processivity factor, is a key component of the active replication fork. It is known to interact with two of three polymerases that localize to the replication fork, DNA polδ and DNA polε (30). Several transacting proteins that can stall or terminate replication fork movement in response to DNA damage have been identified. Biochemical and genomic approaches have identified DNApolα-primase, MCM2–7, Cdc45, MCM10, GINS (Go Ichi, Nii, San), Mrc1, Tof1, and Rrm3 proteins among others at stalled eukaryotic replication forks (31–36). The MCM2–7 complex is the putative replicative helicase that travels with replication forks.

Follicle cell gene amplification is particularly advantageous for studying replication fork progression because of the well-defined domains of replication. The gradient of amplified DNA for all four DAFCs extends ≈50 kb to either side of the maximally amplified DNA at the origin. The extent of the gradient is the same for all of the DAFCs despite the difference in the number of rounds of replication initiation, with DAFC-30B only 4-fold amplified and DAFC-66D amplifying 60- to 80-fold (1, 2, 37, 38). It is not clear whether replication fork barriers (RFBs) limit the size of the gradients and block fork elongation. The existence of RFBs is suggested by the asymmetrical sides of the amplification gradients for DAFC-7F and 66D, in which one side drops off more sharply from the peak (1, 9).

Replication initiation and elongation are developmentally regulated at the DAFC-66D follicle cell amplicon, facilitating analysis of elongation (9). During stages 10B and 11 of egg chamber development sequential replication initiation events and replication elongation occur concurrently, and BrdU-labeled foci are detected at DAFC-66D. During stages 12 and 13, when replication elongation occurs in the absence of further replication initiation events, BrdU-labeling at DAFC-66D resolves into a double-bar structure (Fig. 1A). The double-bar structure reflects BrdU-labeling at replication forks that have traveled away from the single major origin of replication, with the gap distance between bars reflecting the distance traveled by replication forks after initiation (9). The ORC colocalizes with BrdU during the initiation stages, but in stage 11 it is flanked by BrdU, and it is no longer present in stages 12 and 13. In contrast, proteins that travel with the replication fork such as the MCM complex continue to localize with the BrdU labeling, resolving to double bars as well in stages 12 and 13 (9).

Fig. 1.

The distance between BrdU-labeled double bars in cyclinE^1f36/cyclinE^Pz8 mutant follicle cells indicates increased replication fork progression. (A) During gene amplification at DAFC-66D, multiple, tandem replication forks carry out bidirectional replication from a single major origin of replication. BrdU incorporation by replication forks in stage 13 egg chambers produces a double-bar structure (red shading). The bracketed line indicates the gap distance between BrdU-labeled bars, reflective of replication fork progression. (B) cyclinE^1f36/cyclinE^Pz8 mutant follicle cells display BrdU-labeled double bars (red) at DAFC-66D visualized by FISH (green). (Scale bar: 2 μm.) DAFC-66D can be distinguished also from the other DAFCs because it undergoes the highest levels of amplification and has the strongest BrdU signal. (C–F) cyclinE^1f36/cyclinE^Pz8 follicle cells display BrdU-labeled double bars with larger gap distances relative to wild-type CantonS follicle cells. Deconvolved confocal images of BrdU-labeled double bars (C and E) were rendered into 3D projections (D and F) to accurately measure gap distances. (C and D) A CantonS double bar with a gap distance of 0.3 μm. (E and F) A cyclinE^1f36/cyclinE^Pz8 double bar with a gap distance of 0.7 μm.

We have isolated a mutant displaying the previously undescribed phenotype of increased replication fork progression during gene amplification, and strikingly the mutation is in the cyclinE gene. cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 follicle cells display increased replication fork progression, indicating a lack of stringent RFBs and implicating CyclinE in the control of replication fork progression.

Results

Replication Forks Travel Farther in cyclinE^1f36 Mutants.

The pattern of BrdU labeling in a previously undescribed cyclinE mutant during amplification indicated that replication forks moved farther from the origin than in wild-type follicle cells. This allele, cyclinE^1f36, is an EMS allele that was isolated from a screen for regulators of the G₁/S transcriptional program (11, 39). This allele is lethal in trans to Df(3L)TE35D1 that deletes cyclinE and cyclinE^Pz5, a P-element allele. In trans to cyclinE^Pz8, another P-element allele, cyclinE^1f36 is sterile. In trans to cyclinE^P28, a hypomorphic EMS allele, cyclinE^1f36 is semisterile. cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 mutant follicle cells showed double bars of BrdU staining in stage 13 egg chambers with increased gap distances relative to a wild-type control (Fig. 1 C and E). We confirmed that the double bars corresponded to gene amplification at DAFC-66D in cyclinE^1f36/cyclinE^Pz8 mutant follicle cells by performing BrdU/FISH colabeling using a DAFC-66D specific FISH probe (Fig. 1B).

To quantify gap distances in cyclinE^1f36/cyclinE^Pz8, we deconvolved stacked confocal images, rendered 3D projections, and measured the distance between bars (Fig. 1 D and F). The average gap distance for wild-type follicle cells was 0.38 ± 0.06 μm in contrast to 0.54 ± 0.13 μm for cyclinE^1f36/cyclinE^Pz8 (Table 1). Wild-type follicle cells showed gap distances ranging from 0.30 to 0.50 μm, whereas cyclinE^1f36/cyclinE^Pz8 mutant follicle cells had gap distances ranging from 0.30 to 0.80 μm (Table 1). This increase in gap distances between bars in cyclinE^1f36/cyclinE^Pz8 mutant follicle cells is consistent with increased replication fork progression.

Table 1.

Gap distances of double bars in cyclinE^1f36/cyclinE^Pz8 and wild-type (CantonS) follicle cells

Follicle cell no.	BrdU bar separation, μm
	CantonS	cyclinE1f36/cyclinEPz8
1	0.3	0.8
2	0.5	0.6
3	0.4	0.6
4	0.4	0.3
5	0.4	0.7
6	0.4	0.6
7	0.4	0.3
8	0.3	0.5
9	0.3	0.3
10	0.4	0.7
Mean ± SD	0.38 μm ± 0.06	0.54 μm ± 0.13

A t test shows the difference in these mean distances is significant (P < 0.012).

DNA Copy Number in Amplification Gradients Shows Increased Replication Fork Movement in the Mutant.

Bidirectional replication by tandem replication forks produces a gradient of DNA copy number for each DAFC, with the highest DNA copy numbers at the site of the replication origin and decreasing DNA levels to either side. Because follicle cells are terminally differentiated and do not undergo divisions or rearrangement of the amplified DNA, the amplification profile is reflective of the number of replication forks that have passed through and replicated a given site. Despite the fact that for some DAFCs the peak of amplification is 60- to 80-fold, whereas for others it is only 4-fold, the gradient extends over 100 kb in each DAFC (1, 2). Thus the extent of the gradient reflects the progression of the replication forks during amplification. If the increased gap between BrdU double bars in cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 mutant follicle cells resulted from increased replication fork movement, two changes in the amplification gradient were predicted. First, the copy number of DNA in regions flanking the origin would be increased. Second, the span of the amplified region would be increased.

We determined amplification profiles at DAFC-66D by performing quantitative real-time PCR on genomic DNA purified from stage 13 egg chambers and determined fold amplification at DAFC-66D in 10-kb intervals (Fig. 2A). Consistent with increased replication fork progression, in cyclinE^1f36/cyclinE^Pz8 stage 13 egg chambers DNA copy number in regions 20 kb and farther flanking the DAFC-66D origin were elevated. In addition, the amplified region in cyclinE^1f36/cyclinE^Pz8 stage 13 egg chambers extended 100 kb to either side of the amplified maximum, in contrast to the 50 kb observed in the wild-type control (Fig. 2A). We found the same for cyclinE^1f36/cyclinE^P28 (data not shown). The amplification levels at the peak were not significantly different between mutant and wild type by an unpaired student t test (two-tailed P value of 0.13 for the −10-kb value and 0.08 for the 0 value; N of 3). In DNA isolated from FACS purified 16C follicle cells the peak levels of amplification in the cyclinE^1f36/cyclinE^Pz8 and Oregon R control were 20- and 21-fold, respectively (not significantly different with a P value of 0.94).

Fig. 2.

cyclinE^1f36/cyclinE^Pz8 mutant follicle cells display increased DNA copy number at DAFC-66D, consistent with increased replication fork progression. (A) Quantitative real-time PCR was used to determine fold amplification at DAFC-66D for wild-type OregonR (pink) and cyclinE^1f36/cyclinE^Pz8 stage 13 egg chamber DNA. Fold amplification was determined at 10-kb intervals surrounding the amplified maximum at 0 kb. Two independently isolated cyclinE^1f36/cyclinE^Pz8 DNA preparations (1, blue, and 2, yellow) were used. Errors are the standard deviation of the sample. (B) Comparative genomic hybridization of the Drosophila 2L tiled array was performed, using 16C follicle cell DNA and diploid embryonic DNA as templates for probe synthesis. cyclinE^1f36/cyclinE^Pz8 follicle cells (dashed line) displayed increased copy numbers in regions flanking the amplified maximum relative to an OregonR wild-type control (solid line). Plots were smoothened by taking the moving average of 15 contiguous data points. (C) cyclinE^1f36 shows a semidominant replication phenotype. Quantitative real-time PCR of amplification at DAFC-66D for cyclinE^1f36/CyO (yellow), cyclinE^P28/CyO (pink), cyclinE^Pz8/CyO (teal), and Df(2L)TE35D1/CyO (brown) heterozygous mutant stage 13 egg chamber DNAs. OregonR was used as the wild-type control (black). CyO is a balancer chromosome with a wild-type copy of the cyclinE gene. cyclinE^1f36/CyO showed a marked increase in DNA copy number in the region between 20 and 40 kb from the amplified maximum relative to all other samples. (D) Sequencing of cyclinE^1f36 revealed a Gly to Glu substitution (arrow) in a nonconserved residue near the conserved MRAIL substrate recognition motif. Asterisks indicate amino acids conserved between the species shown, and dots indicate conservative changes.

As an independent measure of copy number at DAFC-66D, we performed comparative genomic hybridization (CGH), using the Drosophila 2L tiled array, which also has a 160-kb genomic segment surrounding DAFC-66D on Chromosome 3 (40). We observed increased copy number in regions flanking the amplified maximum (Fig. 2B). The DAFC-30B amplicon also had increased copy number in an expanded amplified region [supporting information (SI) Fig. 5].

The Replication Phenotype in cyclinE^1f36 Is Semidominant.

We sequenced the coding region of cyclinE^1f36 and identified a glycine to glutamate amino acid substitution mutation not present in control strains recovered from the same screen (Fig. 2D). This substitution occurs in the fifth residue N-terminal to the MRAIL hydrophobic patch, which is required for Cyclin binding to substrates containing the RXL motif, including replication factors and CIP/KIP family Cyclin/Cdk inhibitors (41, 42).

To determine whether increased replication fork progression is recessive or dominant, we performed quantitative real time PCR on stage 13 egg chamber DNA, measuring DNA copy number at specific intervals along the DAFC-66D amplicon for cyclinE^1f36/+, cyclinE^Pz8/+ and cyclinE^P28/+ egg chambers (Fig. 2C). These values were compared with those from Df(2L)TE35D1/+, which lacks one copy of cyclinE. Df(2L)TE35D1/+ stage 13 egg chambers displayed a small increase in fold amplification relative to the OregonR wild-type stage 13 egg chambers at 20 kb, 30 kb, 40 kb, and 50 kb from the amplification maximum (Fig. 2C). cyclinE^Pz8/+ displayed an amplification profile very similar to OregonR, and cyclinE^P28/+ follicle cells displayed an amplification profile very similar to Df(2L)TE35D1/+ follicle cells (Fig. 2C). cyclinE^1f36/+ follicle cells, however, displayed considerable increases in fold amplification relative to wild type at 20 kb, 30 kb, and 40 kb (Fig. 2C). The levels were increased to a lesser extent at the 50-kb and 60-kb positions; these levels were lower than in cyclinE^1f36/cyclinE^Pz8 (Fig. 2A), so the cyclinE^1f36 mutation is semidominant rather than dominant. Given the amplification levels observed in the cyclinE deficiency heterozygote, the properties of the cyclinE^1f36 allele are most simply explained by it causing a dominant gain-of-function. It is possible, however, that the CyclinE^1f36 protein form antagonizes the wild-type protein, making levels of functional protein lower than in the deficiency heterozygote, so that the phenotype arises from loss of function. In follicle cells the staining intensity of the MPM2 antigen has been correlated with levels of active CyclinE (7). We attempted to use this MPM2 staining to assess CyclinE activity in the cyclinE^1f36 mutant, but did not observe detectable differences in MPM2 staining intensity compared with wild type (data not shown).

The Localization of Replication Fork Proteins Is Consistent with Increased Fork Movement in the cyclinE^1f36 Mutant.

An alternative explanation for the amplification profiles observed in cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 mutants is that additional origins in the vicinity of the follicle cell amplicons are activated. Firing of these origins adjacent to the normal DAFC-66D origin could increase DNA copy number in flanking regions. We used Drosophila Unigene cDNA microarrays to test for increases in DNA copy number at previously undescribed sites in cyclinE^1f36/cyclinE^Pz8 follicle cells (2), which would suggest that the mutant amplification profile was due to misregulated origin activation, and failed to detect ectopic origin activation (data not shown). Analysis of DNA copy number along chromosome 2L showed no copy number increases other than known amplicons in the cyclinE^1f36/cyclinE^Pz8 follicle cells, also indicating that random activation of origins does not occur in the mutant (SI Fig. 6). In addition, the increased gap distance between double bars is consistent with increased replication fork progression but not additional origin firing. To investigate further whether the expanded amplification profiles observed for cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 follicle cells were due to nonspecific origin usage, we examined the localization of replication factors that have a characterized, dynamic localization pattern to DAFC-66D during gene amplification.

In wild-type follicle cells, ORC localizes as foci specifically to the DAFCs at stage 10B, the developmental period during which replication initiation events occur (8, 11). The ORC foci dissipate during stage 11 and by stage 12, when only replication elongation occurs at DAFC-66D, ORC no longer localizes to DAFC-66D (9, 11). If the increased DNA copy number in amplicons from mutant follicle cells were the consequence of activation of origins in flanking sequences, one possibility would be persistent ORC localization to foci in stages 11 and 12. Staining with antibodies against the ORC2 subunit showed that ORC localized to specific foci in both wild-type and cyclinE^1f36/cyclinE^Pz8 mutant follicle cells in stage 10B (Fig. 3 A and C). In stage 11 and later, ORC localization to the DAFCs diminished in both wild-type and the cyclinE mutants (Fig. 3 B and D). This disappearance indicates that replication initiation events are confined to the appropriate developmental period in cyclinE^1f36/cyclinE^Pz8 mutant follicle cells. In addition, we were unable to detect additional ORC foci, consistent with replication initiation not occurring from ectopic origins in cyclinE^1f36/cyclinE^Pz8 mutants.

Fig. 3.

Localization of replication factors in cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 follicle cells. In each panel, a single follicle cell nucleus is shown, and in two-color images, DNA is shown in green and antibody staining in red. ORC2 localization in cyclinE^1f36/cyclinE^Pz8 (C and D) and CantonS wild-type (A and B) follicle cells. (A and B) In CantonS wild type, ORC2 localizes to two major foci in stage 10B (A). One of these ORC2 foci corresponds to replication initiation events at DAFC-66D (9). (B) In stage 11, ORC2 delocalizes from foci concurrent with cessation of replication initiation events at DAFC-66D. (C and D) In cyclinE^1f36/cyclinE^Pz8 follicle cells, ORC2 localizes to two major foci in stage 10B (C). (D) In stage 11, ORC2 delocalizes from foci in cyclinE^1f36/cyclinE^Pz8 follicle cells, suggesting that replication initiation events are appropriately restricted. (E–H) MCM2–7 localization to double bars in OregonR and cyclinE^1f36/cyclinE^Pz8 follicle cells in stage 13 egg chambers. (I–L) DUP localization to double bars in OregonR wild-type and cyclinE^1f36/cyclinE^Pz8 follicle cells in stage 13 egg chambers. (M–P) PCNA localization to double bars in CantonS and cyclinE^1f36/cyclinE^P28 follicle cells in stage 13 egg chambers. Single confocal images or projections of stacked confocal images are shown to visualize double bars (E–P). ORC2 localization was visualized as projections of stacked confocal images.

The replication factors DUP (Drosophila Cdt1), the MCM complex and PCNA localize to replication forks during gene amplification (9). During stages 10B and 11, when replication forks are proximal to the origin and being actively generated at the origin, DUP, the MCM complex and PCNA localize to foci at DAFC-66D (9). By stage 13, replication forks generated by replication initiation events during stages 10B and 11 have moved away from the DAFC-66D replication origin and concomitantly, DUP, the MCM complex and PCNA colocalize with BrdU as double bars at DAFC-66D (9).

If the expanded amplification profiles of cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 mutants were due to increased replication fork progression, we would expect to see double-bar staining with increased gap distances for these proteins in the mutant follicle cells. We saw MCM localized, at reduced levels, to double bars with dramatically increased gap distances, relative to an OregonR wild-type control (Fig. 3 E–H). In addition, in cyclinE^1f36/cyclinE^Pz8 follicle cells, we saw DUP localized, at reduced levels, to double bars (Fig. 3 K and L). Although gap distances in these double bars seemed to be increased in cyclinE^1f36/cyclinE^Pz8 follicle cells, gap distances were not increased as dramatically as seen for MCM double bars in cyclinE^1f36/cyclinE^Pz8 follicle cells. The reduced amount of MCM and DUP staining at the DAFCs in the cyclinE^1f36/cyclinE^Pz8 mutant cells could reflect an involvement of CyclinE/CDK2 in localizing these replication factors. PCNA antibody staining was not as clear, and it was difficult to observe PCNA localized to double bars in wild type, but double-bar staining could be seen in cyclinE^1f36/cyclinE^P28 follicle cells (Fig. 3 M–P). The resolution of DUP, MCM2–7 and PCNA by immunofluorescence into double bars in stage 13 egg chambers in cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 follicle cells is consistent with increased replication fork progression.

The localization of DUP, MCM2–7 and PCNA shows that these replication factors are found at replication forks localized at the ends of the amplified domains and not spread throughout the amplicon. This pattern is consistent with the absence of ORC localization in late stage egg chambers, supporting the conclusion that replication forks move farther and not that new origins fire in flanking regions. In addition, these studies show that levels of DUP, the MCM complex, and PCNA are not detectably increased, excluding increased levels as an explanation for increased replication fork progression in the cyclinE^1f36 mutants.

Increased Replication Fork Progression Is Not Due to Prolonged Egg Chamber Development.

The follicle cells are terminally differentiated, and they are sloughed off as the egg passes into the uterus. Replication forks move slowly in the follicle cell amplicons, as they do in other polytene tissues (9, 43). The gradients of amplification observed in wild-type follicle cells can be accounted for by replication forks initiated in stage 10 and 11 moving at a constant rate until the follicle cells die in stage 14 (9). Thus, one possibility was that increased replication fork progression and gradient width in the cyclinE mutant were due to a developmental delay during oogenesis, rather than a direct effect on fork movement rate or processivity. We assayed egg-laying rates over twelve days to determine the time required for oogenesis in cyclinE^1f36/cyclinE^Pz8 females relative to sibling control females. We determined that the developmental timing of oogenesis is indistinguishable between cyclinE^1f36/cyclinE^Pz8 and heterozygous sibling ovaries (data not shown), revealing that increased replication fork progression in cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 mutants is not due to developmental delays in oogenesis.

Gene amplification at DAFC-66D initiates during follicle cell endocycles, to result in ≈4-fold amplification by stage 10A (7). Premature extra rounds of gene amplification during endocycle S-phases in cyclinE^1f36/cyclinE^Pz8 follicle cells would provide longer developmental times for replication fork progression. We excluded this possibility by showing that the DNA copy number increases in the flanking portions of the gradient for DAFC-66D in fact occur between stages 11 and 13, and thus are due to replication fork movement. The copy number changes indicate that replication forks travel 40–60 kb between stages 11 and 13 (Fig. 4A). In contrast, previous work showed that in wild-type OregonR, replication forks travel only 10–20 kb between stages 11 and 13 (9).

Fig. 4.

Replication forks progress farther at DAFC-66D and DAFC-7F in cyclinE^1f36/cyclinE^Pz8 follicle cells during the elongation phase of amplification. (A) Real-time PCR quantification of DNA copy number at DAFC-66D using DNA isolated from staged cyclinE^1f36/cyclinE^Pz8 egg chambers showed that replication forks move ∼60 kb between stages 11 (pink) and 13 (yellow) (arrows). (B) Quantification of BrdU-labeled foci in cyclinE^1f36/cyclinE^Pz8 (red) and sibling control (blue) stage 12 egg chambers demonstrated that cyclinE^1f36/cyclinE^Pz8 follicle cells have more BrdU-labeled foci, reflective of active replication forks, than sibling control follicle cells (P < 0.001). (C) Quantitative real-time PCR was performed on staged egg chambers to assay gene amplification at DAFC-7F in cyclinE^1f36/cyclinE^Pz8 and wild-type OregonR ovaries. At 30 kb from the amplified maximum, cyclinE^1f36/cyclinE^Pz8 stage 13 egg chambers (pink) show increased copy number relative to the OregonR control (yellow), consistent with replication forks traveling farther during this period. Stage 11 egg chamber copy number is shown for cyclinE^1f36/cyclinE^Pz8 (blue) and OregonR (teal). (D) BrdU-labeled double bars flanking a DAFC-7F FISH signal were observed in cyclinE^1f36/cyclinE^Pz8 mutants, but they are not seen in wild type, also consistent with increased replication fork progression at DAFC-7F. Arrow points to DAFC-66D double bar and arrowhead points to DAFC-7F double bar. In the merged image, BrdU staining is in red and the FISH signal in green.

The effect of the cyclinE^1f36 mutant on other amplicons also supports the conclusion that the altered amplification gradients are due to increased replication fork movement. In cyclinE^1f36/cyclinE^Pz8 mutant follicle cells there is increased gradient width in at least two other amplicons, DAFC-7F and DAFC-30B, and these DAFCs do not initiate amplification until stage 10, after the cessation of endocycles (2, 7). We observed an increased number of BrdU foci in stage 12 cyclinE^1f36/cyclinE^Pz8 mutant follicle cells (Fig. 4B). This result could be consistent with either the presence of more separated replication forks or the firing of additional origins. We distinguished these possibilities by quantification of DNA copy number between stages 11 and 13 at DAFC-7F and showed that in this mutant replication forks move farther than in wild type (Fig. 4C). This increased distance can be seen cytologically. Normally in wild-type follicle cells in stage 12 egg chambers, BrdU foci corresponding to replication at DAFC-7F are not visible. In contrast, in the cyclinE mutant, BrdU double bars, indicating more incorporation of nucleotides and further separation for replication forks, are seen at DAFC-7F (Fig. 4D). This result suggests that the additional BrdU foci observed (Fig. 4B) are caused by more separated replication forks rather than initiation from additional origins. Thus, CyclinE affects replication fork progression at multiple DAFCs.

Discussion

We have shown that cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 follicle cells display increased replication fork progression. We observed that replication forks at DAFC-30B and DAFC-66D move twice as far in the mutants, and we also found increased progression at DAFC-7F. Thus this effect is not restricted to one genomic interval. The cyclinE^1f36 allele is unusual in that it is highly specific to replication elongation. In cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 follicle cells, DAFC-66D is amplified (at maximum) to levels comparable with wild type, and the cell cycles leading up to gene amplification are minimally perturbed. We did not detect delays in oogenesis in cyclinE^1f36/cyclinE^Pz8 ovaries, demonstrating that increased replication fork progression in cyclinE^1f36/cyclinE^Pz8 follicle cells is not due to a prolonged period between the onset of gene amplification and the sloughing off of follicle cells in stage 14 egg chambers.

Our results are consistent with an absence of site-specific termination mechanisms for replication forks during amplification. The extension of amplicons to 200 kb in the cyclinE mutants indicates that there are not RFBs that normally limit replication fork progression to 50 kb from the origin. The only way such barriers can be reconciled with our results would be if CyclinE were essential to site-specific RFB function during gene amplification, for example by regulating transacting factors. It seems more likely that replication fork rate or processivity in combination with the elapsed time between replication initiation events of gene amplification (occurring largely in stages 10B and 11 at DAFC-66D) and follicle cell removal in stage 14, determines the span of gene amplification. By accelerating replication fork movement or enhancing fork processivity the cyclinE mutation produces larger amplified domains.

Increased replication fork progression is a semidominant phenotype conferred by cyclinE^1f36. This mutation was identified initially because it altered the transcriptional program normally affecting the introduction of a G₁ phase to the cell cycle in embryogenesis (39). This effect may be due to independent activities of the CyclinE/Cdk2 complex affected by the cyclinE^1f36 mutation. We did not observe consequences for replication elongation of eliminating one copy of the cyclinE gene, suggesting that it is not a reduction of CyclinE function that affects fork progression but activities modified by this mutation. We cannot exclude the possibility, however, that the mutant form of CyclinE protein affects the wild-type protein in a dominant-negative manner. We did not detect appreciable changes in CyclinE protein levels by whole-mount antibody staining of cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 ovaries (data not shown). This mutant allele carries a Gly to Glu substitution 5 aa removed from the MRAIL hydrophobic patch. The MRAIL hydrophobic patch interacts with cognate RXL sequences and is important for Cyclin/Cdk substrate specificity (44), inhibition of Cyclin/Cdk by CIP/KIP inhibitors (45) and chromatin localization of Clb5 in S. cerevisiae and CyclinE in Xenopus via interaction with RXL motifs in Orc6 and Cdc6, respectively (41, 46). The proximity of the Gly to Glu substitution found in cyclinE^1f36 to the MRAIL hydrophobic patch suggests that this mutation disrupts CyclinE/Cdk2 or CyclinE binding to replication factors. On the 3-D crystal structure of mouse Cyclin E the corresponding amino acid to the mutated Gly in cyclinE^1f36 lies between two alpha helices, and it is not obvious that it would disrupt the structure (47).

There are three mechanisms by which mutation of cyclinE could cause increased fork progression: (i) a direct effect on the activity of proteins at the replication fork; (ii) a change in chromatin structure that facilitates movement of the replication fork; or (iii) a disruption of polytene chromosome structure, as it is known that replication forks progress slowly in polytene cells. Note that all of these effects could act to increase rates of replication fork movement or enhance processivity, and our results do not distinguish between increased rates of fork movement and increased processivity. This role for CyclinE may not be restricted solely to amplification. Based on observations that CDK2 is recruited to chromatin in a open configuration and participates in its decondensation, Alexandrow and Hamlin speculated that CyclinE could facilitate fork movement (48).

CyclinE may directly regulate replication fork progression by covalently modifying replication fork components. In particular, an appealing model is that the cyclinE^1f36 phenotype is due to altered CyclinE/Cdk2 regulation of DUP/Cdt1 or the MCM2–7 complex. DUP/Cdt1 travels with the replication forks during amplification, is required for MCM loading, and has been shown by a dominant negative form to be required for elongation (9, 49). cyclinE^1f36 could also act directly on the MCM complex to promote its activity. This effect would most likely be via a novel CyclinE activity, as phosphorylation inhibits MCM4,6,7 subcomplex helicase activity (50). Biochemical studies on the CyclinE^1f36 protein will be necessary to determine whether it directly affects DUP or the MCM complex.

It is also possible that chromatin configuration limits replication fork movement. CyclinE genetically interacts with proteins known to affect chromatin structure (51), and thus it is conceivable that in the cyclinE^1f36 mutant the chromatin configuration is altered to be more permissive for replication fork progression. We did not observe histone acetylations associated with replication forks during amplification (52), so it is unclear what modifications or binding proteins would change the accessibility of chromatin for fork movement (53).

Increased replication fork progression might result from disrupted polytene structure. Polyteny affects replication fork speed. Measurements of replication fork speed indicate that whereas replication forks move at ≈2.6 kb/min in Drosophila diploid cells, replication fork speed is an order of magnitude lower in Drosophila polytene larval salivary glands at ≈300 bp/min (54). This order of magnitude difference may be due to protein/DNA complexes specific to polytene chromosomes or the structure of polytene chromosomes themselves. Before gene amplification, follicle cells undergo endocycles resulting in polytene chromosomes, which are maintained during gene amplification (55). The polytene structure of follicle cell chromosomes could serve as a major determinant of replication fork progression during gene amplification. Thus we measured polytene chromosome structure in the cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 follicle cells by FISH hybridization and concluded that altered fork movement is not due to disrupted polyteny (SI Fig. 7).

The functional significance of actively restricting replication fork progression during gene amplification is yet unclear. We examined the eggshells of eggs laid by cyclinE^1f36/cyclinE^Pz8 females by Scanning Electron Microscopy. Follicle cell footprints, dorsal appendages, micropyle and operculum looked completely normal, indicating that increased replication fork progression does not disrupt eggshell formation (SI Fig. 8). Although we were unable to detect eggshell defects, cyclinE^1f36/cyclinE^Pz8 and cyclinE^1f36/cyclinE^P28 females were sterile and semisterile respectively, suggesting that limiting replication fork progression during follicle cell gene amplification may be important for egg viability. An intriguing possibility is that misregulated replication fork progression disrupts the expression of genes important for egg viability by impeding transcription or altering copy number.

Materials and Methods

Stocks.

The cyclinE^1f36 allele was isolated from an ethylmethanesulfonate mutagenesis screen for mutations affecting the G₁-S transition and shown to fail to complement cyclinE (39). The cyclinE^Pz8 and cyclinE^P28 alleles were provided by Helena Richardson (56) and Df(3L)TE35D1 was from John Roote. Either CantonS or OregonR were used as wild-type strains. Strains were grown under standard conditions.

BrdU Labeling and Immunostaining.

BrdU labeling and anti-ORC2, anti-DUP, anti-MCM2–7, and anti-PCNA antibody staining were performed on whole ovaries as described (9). YOYO-1 (Molecular Probes, Eugene, OR) was used to stain DNA at 1:2,000. BrdU/FISH colabeling was performed as described (2). Imaging was performed by using a Zeiss Axiovert 100 M Meta confocal microscope with LSM510 software. Deconvolution was carried out by using Huygens2.3-professional (Scientific Volume Imaging, Hilversum, The Netherlands). Rendering and analysis of three-dimensional data were carried out by using the MeasurementPro module of Imaris3 Surpass 3.2 (Bitplane, St. Paul, MN). The gap distances between BrdU bars were measured between the inner surfaces, measured at multiple positions along the length to ensure the shortest distance was measured. Measurements of the gap distances were sensitive to 0.1 μm.

Quantitative Real-Time PCR.

Quantitative real-time PCR was performed by using primer sets spanning 50 kb on either side of ACE3 (denoted as distance 0) at 10 kb intervals and primers to a nonamplified intergenic region on chromosome arm 3R (located ≈25 kb upstream of the DNApolα locus) as described (9). Primers used for DAFC-7F were from (57). Primer sets spanning 60–100 kb on either side of ACE3 were generated as described (9) and supplied by GeneLink (Hawthorne, NY). Three 10-fold dilutions of stage 1–8 egg chamber DNA were used as standards. Relative fluorescence was measured for each sample in relation to the standard curves, and standard deviations of triplicate reactions were calculated by the ABI Prism 7000 software. Fold amplification was calculated by dividing the relative fluorescence of the amplicon site by the relative fluorescence for the nonamplified intergenic region near DNApolα. Error is the standard deviation of the ratio A/C = (FA/FC)[(SA/FA)² + (SC/FC)²] (A, amplicon locus; C, control locus; FA, relative fluorescence for amplicon locus, FC, relative fluorescence of control locus; SA, standard deviation for amplicon locus; SC, standard deviation for control locus).

Drosophila 2L Tiling Array.

The microarrays have been described and contain 1.5-kb PCR products tiling chromosome 2L and covering the region on 3L with DAFC-66D (40). DNA from 16C follicle cells and embryonic DNA for reference were purified as described (2). DNA was labeled, slides were hybridized, and data were analyzed as described (40). To calculate P values, total data points were used to model the normal distribution. Experiments were performed in triplicate with one dye-swap.

Sequencing.

Sequencing was performed as described by using genomic DNA from eight embryos for each sequencing run (58). cyclinE coding regions were sequenced with 2-fold coverage, and the mutation was sequenced with 4-fold coverage. Another strain recovered in the same screen, thus derived from the same isogenized chromosome (39), was sequenced in parallel, and the Gly to Glu substitution was not present in this strain.

Abbreviations

ORC

origin recognition complex

DAFC

Drosophila amplicon in follicle cell

PCNA

proliferating cell nuclear antigen

ACE

amplification control element

RFB

replication fork barrier

MCM

minichromosome maintenance

DUP

double parked.